AI News AI资讯 18h ago Updated 14h ago 更新于 14小时前 48

Anthropic changes data retention policy after enterprise pushback Anthropic因企业反对而更改数据保留政策

Anthropic is revising its controversial 30-day data retention policy for enterprise customers using Mythos and Fable models Customer data will now be stored in the customer's own cloud infrastructure rather than Anthropic's servers, while maintaining the 30-day retention window The change follows significant enterprise pushback, with Anthropic acknowledging the original policy was unpopular and a business risk Over 100 customers from regulated industries were consulted during the months-long dev Anthropic调整数据保留政策,回应企业客户对数据隐私的强烈反对 30天数据存储窗口保留,但数据将存储在企业自有云端而非Anthropic服务器 新方案与100多家受监管行业客户合作开发,预计秋季上线 OpenAI也在探索类似的安全与数据控制结合方案

72
Hot 热度
65
Quality 质量
68
Impact 影响力

Analysis 深度分析

TL;DR

  • Anthropic is revising its controversial 30-day data retention policy for enterprise customers using Mythos and Fable models
  • Customer data will now be stored in the customer's own cloud infrastructure rather than Anthropic's servers, while maintaining the 30-day retention window
  • The change follows significant enterprise pushback, with Anthropic acknowledging the original policy was unpopular and a business risk
  • Over 100 customers from regulated industries were consulted during the months-long development of the new system
  • The updated policy is expected to roll out this fall, with rival OpenAI pursuing a similar security-plus-control approach via partnerships with Databricks and Microsoft

Why It Matters

This policy shift highlights the growing tension between AI safety monitoring and enterprise data privacy, a critical concern for any organization handling sensitive or regulated information. It signals that major AI providers are recognizing data governance as a competitive differentiator, not just a compliance checkbox. The move also reflects increasing pressure on AI companies to align their operational practices with the stringent requirements of regulated industries.

Technical Details

  • The original policy stored all customer data from Mythos, Fable, and future flagship models on Anthropic's servers for 30 days, primarily for cybersecurity monitoring to detect novel attacks leveraging their technology
  • The revised architecture shifts data storage to customer-controlled cloud environments while preserving the 30-day retention window, effectively decoupling Anthropic from direct data custody
  • The new system was co-developed with over 100 enterprise customers across regulated sectors, suggesting a focus on industries with strict data sovereignty and compliance requirements (e.g., finance, healthcare, government)
  • OpenAI is pursuing a parallel but distinct approach through partnerships with Databricks and Microsoft, indicating the industry is exploring multiple models for balancing security oversight with data control

Industry Insight

  • Data sovereignty will become a key purchasing criterion for enterprise AI adoption; companies that offer flexible data residency options will gain a competitive edge in regulated markets
  • AI providers should expect ongoing pushback on data retention policies and proactively design systems that give customers granular control over their data lifecycle
  • The convergence of security monitoring and data governance solutions (as seen with both Anthropic and OpenAI) suggests a new category of AI infrastructure partnerships will emerge, combining cloud providers, data platforms, and model vendors

TL;DR

  • Anthropic调整数据保留政策,回应企业客户对数据隐私的强烈反对
  • 30天数据存储窗口保留,但数据将存储在企业自有云端而非Anthropic服务器
  • 新方案与100多家受监管行业客户合作开发,预计秋季上线
  • OpenAI也在探索类似的安全与数据控制结合方案

为什么值得看

Anthropic的政策调整反映了企业级AI服务中数据安全与模型安全之间的核心矛盾,标志着云厂商开始重新平衡安全监控与客户数据主权。这一变化对AI从业者意味着企业客户的数据控制权正在成为云服务提供商必须解决的关键问题。

技术解析

  • 数据存储架构变更:从Anthropic自有服务器迁移至客户云端,保持30天安全分析窗口
  • 合作开发模式:与100多家受监管行业客户共同构建新系统
  • 安全与隐私平衡:在保留网络攻击检测能力的同时满足企业数据主权需求
  • 竞品动态:OpenAI与Databricks、Microsoft合作探索类似方案

行业启示

  • 企业级AI服务的数据主权将成为竞争关键,云厂商需重新设计数据存储架构
  • 受监管行业对AI服务的合规要求将推动行业标准的建立
  • AI安全研究与企业数据隐私的平衡需要技术创新而非简单妥协

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Claude Claude Security 安全 Policy 政策